By Thomas O'Kelly, Senior Research Manager
Why perception, not reality, is the thing social research has to measure
Ask people across Western Europe whether crime is rising, and the answer is close to unanimous. New YouGov polling across six countries finds large majorities convinced that crime has gone up in recent years. In Italy the figure reaches 80%. In France it is 78%. In Britain it is 66%. Only Denmark slips below a majority, and even there more than half believe things are getting worse.
The data tells a different story. In England and Wales the Crime Survey is the most reliable measure available, because it is built on what people actually experience rather than on what police choose to record. It shows that violence, burglary and car crime have fallen by close to 90% over thirty years. This pattern holds across most high-income countries and represents a broad scientific consensus that the long fall in crime is real.
For communicators, this should not be seen simply as the public being wrong, but as perhaps the most important thing to understand. People change their behaviour, decide whom to trust and support policies based on the world as they perceive it, not always as it truly is. Understanding what creates that gap, what it reveals and what it distorts is therefore essential to designing communication that can genuinely change minds.
The country is dangerous. My street is fine.
The clearest fingerprint of the perception gap appears the moment you ask people to look closer to home. In the same YouGov study, the numbers drop significantly when the question shifts from the nation to the respondent's own area. In Britain, 66% think crime in general has risen nationally, but only 53% feel it where they live. For violent crime the split is wider still, 59% nationally against 42% locally. France runs from 78% nationally to 55% locally. Italy from 80 to 61. Every country shows the same shape. The country is going to the dogs, but the street outside is much as it ever was.
This is not a quirk of one survey. Gallup has tracked the same divide in the United States for decades, an effect researchers call the local positivity bias, or more plainly, hometown favouritism. Over that period, Americans have rated crime as worse nationally than locally by an average of 43 percentage points. In one recent reading, 78% thought crime had risen across the country while only 38% felt it had risen where they live. People judge what is in front of them on evidence, and judge everything beyond it on a feeling. The local figure is anchored in lived experience. The national figure is built almost entirely from second-hand impressions.
Why we get it so wrong
If perception of the country comes from impressions rather than experience, the obvious question is where those impressions come from. The answer is the information environment. People do not carry crime statistics in their heads. They carry a running tally of vivid, frightening, recent examples, and they mistake the ease of recalling them for evidence of how common they are. A single shocking case, replayed across news and feeds, weighs more heavily than a national dataset that quietly improves for thirty years. The old newsroom rule, if it bleeds it leads, was a description of this bias long before anyone built an algorithm to exploit it.
Today the bias has industrial scale behind it. Our own analysis of the information environment found that emotionally charged content, particularly anger and fear, was shared up to 365 times more frequently than neutral or positive content. Platforms are tuned to amplify exactly the content that distorts perception most. A reassuring statistic has almost no chance against a distressing clip. The result is a feed that systematically over-represents the rare and the violent, and a public that reasonably concludes the world is more dangerous than it is.
Perception is not about right and wrong
It would be easy to conclude that the public is wrong and the data is right. The truth is more uncomfortable. Perception is not random noise. It is selectively wrong, and the selection tells you something.
Britain stands out in the data for its anxiety about knife crime, named as a distinctive national problem by 60%, far above any other measure. Knife offences did rise, roughly doubling over the decade to 2024, though much of that reflects better recording and numbers remain well under 1% of all recorded crime. The public has gripped onto a real signal and inflated it. Ipsos found that 71% of Britons believe knives cause the most deaths from violence, when the real figure is about a quarter.
This is the pattern beneath the noise. People do not invent their fears. They take a genuine fragment of reality and scale it up. As Ipsos puts it, we overestimate what we worry about as much as we worry about what we overestimate. A misperception is rarely a simple mistake.
What this means for anyone trying to understand a public
If you want to predict how people will vote, what policy they will tolerate, where they will let their children walk, or whether they trust the institutions meant to protect them, real world statistics are pretty useless. What matters is the perception, not the reality, because reality only moves behaviour once it has been filtered through belief. And you cannot fact your way out of a perception gap. A decade of evidence shows that repeating the true number does little to shift the belief, because the belief was never built from numbers. It was built from emotion, memory and prevailing narratives.
So the question for anyone who studies or communicates with the public is not simply how to correct what people get wrong. It is how to understand what those perceptions reveal, what they distort, what behaviours they drive, and what forms of communication they make possible or impossible. In social research, the gap between reality and perception is not a flaw in the data. It is often the thing we need to understand most.
Why do some stories spread while others disappear?
Explore our research into how narratives, emotion and online behaviour shape public perception.
Download our free Information Environments report